[gambit-list] porting to system without tty or standard IO

Scott Graham sgraham at gmail.com
Sat May 28 17:07:35 EDT 2005


Sorry about that, should have been more clear. I was getting a C/OS
level exception, because of an invalid pointer. Since you feel it
should be doable, I suppose it was probably my poor attempts at some
of the os_ functions that were causing the problem.

I tried to do something along the lines of web-repl, but I believe I
still need to at least no-op all standard stream stuff because (for
example) _thread.scm does an (##open-all-predefined), which as far as
I can tell will always get run (or is it only when a thread related
function is used?).

Anyway, thanks for your response. I'll have another try and see if I
can narrow down the exact cause of the exception I was getting.

scott



On 5/28/05, Marc Feeley <feeley at iro.umontreal.ca> wrote:
> On 27-May-05, at 4:35 PM, Scott Graham wrote:
> 
> > Hi
> >
> > Has anyone tried porting Gambit to a system without a tty and/or
> > without the low-level file descriptor level? I'm trying to port to an
> > embedded system that's quite similar to USE_WIN32 as far as os_*.[ch]
> > are concerned, however, it has neither a TTY, nor stdin.
> >
> > I had a try, and got as far as gsi compiling, but I'm getting an
> > exception that I don't understand early on startup.
> >
> 
> Can you be more specific... a Scheme exception, a C exception, a signal?
> Which one?
> 
> It is possible to run Gambit with no stdin/stdout/stderr.  But you have
> to avoid referencing them, and if you just start the interpreter in a
> way
> that will normally start a REPL, the system will raise a Scheme
> exception
> as soon as it tries to display the banner to the user.
> 
> > So, just wondering if anyone has tried this, or has suggestions on
> > plan of attack. Ideally, I'd like to have the system work over a
> > socket connected to the host PC rather than expecting a variety of
> > streams to be available "locally".
> >
> 
> There's an example (in examples/web-repl/web-repl.scm) showing how this
> can be done.
> 
> Marc
> 
>


More information about the Gambit-list mailing list